https://aiweekly.co/alerts/claude-system-prompt-bans-lyrics-after-sony-warner-sue AI Weekly en 2026-09-02T18:12:01+00:00 Claude system prompt bans lyrics after Sony, Warner sue https://aiweekly.co/alerts/axon-draft-one-ai-wrote-texas-sheriffs-flock-abortion-report AI Weekly en 2026-09-02T18:12:01+00:00 Axon Draft One AI wrote Texas sheriff's Flock abortion report https://aiweekly.co/alerts/mrbeast-signs-multi-year-google-deal-for-gemini-fitbit-air AI Weekly en 2026-09-02T16:56:51+00:00 MrBeast Signs Multi-Year Google Deal for Gemini, Fitbit Air https://aiweekly.co/alerts/thyme-care-closes-125m-series-e-doubles-valuation-above-2b AI Weekly en 2026-09-02T16:56:51+00:00 Thyme Care Closes $125M Series E, Doubles Valuation Above $2B https://aiweekly.co/alerts/qwen-drive-10-folds-3d-perception-and-planning-into-one-vlm AI Weekly en 2026-09-02T15:54:39+00:00 Qwen-Drive-1.0 folds 3D perception and planning into one VLM https://aiweekly.co/alerts/one-self-hosted-llm-covers-200-apps-beats-7-baseline AI Weekly en 2026-09-02T15:54:39+00:00 One self-hosted LLM covers 200+ apps, beats ~7× baseline https://aiweekly.co/alerts/burtsev-proposes-ai-threshold-for-self-amplifying-ai-rd AI Weekly en 2026-09-02T15:54:39+00:00 Burtsev Proposes ℜ_AI Threshold for Self-Amplifying AI R&D https://aiweekly.co/alerts/paper-better-safeguard-scores-dont-prove-a-safer-llm-system AI Weekly en 2026-09-02T15:54:39+00:00 Paper: better safeguard scores don't prove a safer LLM system https://aiweekly.co/alerts/hiddenlayer-raises-100m-series-b-as-ai-security-arr-jumps-10x AI Weekly en 2026-09-02T15:26:11+00:00 HiddenLayer Raises $100M Series B as AI-Security ARR Jumps 10x https://aiweekly.co/alerts/google-avoids-adx-sale-as-judge-orders-ad-tech-interoperability AI Weekly en 2026-09-02T15:26:11+00:00 Google Avoids AdX Sale as Judge Orders Ad-Tech Interoperability https://aiweekly.co/alerts/adobe-wires-70-creative-tools-into-slack-via-mcp-app AI Weekly en 2026-09-02T15:26:11+00:00 Adobe Wires 70+ Creative Tools Into Slack via MCP App https://aiweekly.co/alerts/anthropic-launches-free-browser-checker-for-claude-marked-files AI Weekly en 2026-09-02T15:26:11+00:00 Anthropic launches free browser checker for Claude-marked files https://aiweekly.co/alerts/loudouns-data-center-boom-collides-with-homeowner-backlash AI Weekly en 2026-09-02T15:12:01+00:00 Loudoun's data-center boom collides with homeowner backlash https://aiweekly.co/alerts/llms-write-more-similar-stories-than-humans-paper-finds AI Weekly en 2026-09-02T15:12:01+00:00 LLMs write more similar stories than humans, paper finds https://aiweekly.co/alerts/elon-post-poll-27-of-us-adults-turn-to-ai-for-personal-queries AI Weekly en 2026-09-02T13:59:05+00:00 Elon-Post Poll: 27% of US Adults Turn to AI for Personal Queries https://aiweekly.co/alerts/t-techs-merged-qwen3-32b-handles-116m-requests-beats-qwen3-235b AI Weekly en 2026-09-02T13:59:05+00:00 T-Tech's Merged Qwen3-32B Handles 116M Requests, Beats Qwen3-235B https://aiweekly.co/alerts/uber-cuts-3300-roles-thins-managers-20-to-fund-robotaxi-push AI Weekly en 2026-09-02T12:26:56+00:00 Uber Cuts 3,300 Roles, Thins Managers 20% to Fund Robotaxi Push https://aiweekly.co/alerts/navers-dronecats-finds-mllms-fly-fail-to-declare-arrival AI Weekly en 2026-09-02T12:26:56+00:00 NAVER's DroneCATS Finds MLLMs Fly, Fail to Declare Arrival https://aiweekly.co/alerts/dysons-499-camerajet-toothbrush-watches-teeth-with-onboard-ai AI Weekly en 2026-09-02T12:12:01+00:00 Dyson's $499 CameraJet toothbrush watches teeth with onboard AI https://aiweekly.co/alerts/fbi-probes-nexus-sale-of-153m-drivers-licenses-tied-to-idscan AI Weekly en 2026-09-02T10:59:35+00:00 FBI probes Nexus sale of 153M drivers licenses tied to IDScan https://aiweekly.co/alerts/openais-chatgpt-health-plugs-into-epics-patient-charts AI Weekly en 2026-09-02T09:25:33+00:00 OpenAI's ChatGPT Health Plugs Into Epic's Patient Charts https://aiweekly.co/alerts/probasco-ai-risks-hollowing-out-us-military-judgment AI Weekly en 2026-09-02T09:12:01+00:00 Probasco: AI Risks Hollowing Out U.S. Military Judgment https://aiweekly.co/alerts/korea-commits-919b-to-184gw-ai-datacenter-buildout-by-2035 AI Weekly en 2026-09-02T07:56:07+00:00 Korea Commits $919B to 18.4GW AI Datacenter Buildout by 2035 https://aiweekly.co/alerts/smelt-loops-middle-moe-layers-twice-saves-68-18-flops AI Weekly en 2026-09-02T07:56:07+00:00 SMELT loops middle MoE layers twice, saves 6.8-18% FLOPs https://aiweekly.co/alerts/altman-recasts-singularity-as-slow-merger-not-explosion AI Weekly en 2026-09-02T06:12:01+00:00 Altman recasts singularity as slow merger, not explosion https://aiweekly.co/alerts/recurrent-depth-lm-hits-50b-equivalent-compute-at-35b-params AI Weekly en 2026-09-02T06:12:01+00:00 Recurrent-depth LM hits 50B-equivalent compute at 3.5B params https://aiweekly.co/alerts/41-ai-researchers-call-chain-of-thought-monitoring-fragile AI Weekly en 2026-09-02T06:12:01+00:00 41 AI researchers call chain-of-thought monitoring fragile https://aiweekly.co/alerts/ideal-distills-students-without-real-data-matches-imagenet-run AI Weekly en 2026-09-02T06:12:01+00:00 IDeaL distills students without real data, matches ImageNet run https://aiweekly.co/alerts/zimablue-lifts-zero-shot-robot-manipulation-from-361-to-778 AI Weekly en 2026-09-02T03:54:45+00:00 ZimaBlue lifts zero-shot robot manipulation from 36.1% to 77.8% https://aiweekly.co/alerts/ui-venus-2-scales-gui-agent-across-170-multilingual-apps AI Weekly en 2026-09-02T03:54:45+00:00 UI-Venus-2 Scales GUI Agent Across 170+ Multilingual Apps https://aiweekly.co/alerts/smelt-loops-moe-middle-layers-cutting-training-flops-68-180 AI Weekly en 2026-09-02T03:54:45+00:00 SMELT loops MoE middle layers, cutting training FLOPs 6.8-18.0% https://aiweekly.co/alerts/h3-world-turns-minimax-h3-33b-video-model-into-a-world-model AI Weekly en 2026-09-02T03:54:45+00:00 H3-World turns MiniMax-H3 33B video model into a world model https://aiweekly.co/alerts/microsofts-studentsim-doubles-gpt-54-on-chess-learner-fidelity AI Weekly en 2026-09-02T03:26:35+00:00 Microsoft's StudentSim doubles GPT-5.4 on chess-learner fidelity https://aiweekly.co/alerts/willison-puts-claude-fable-51-through-his-pelican-svg-test AI Weekly en 2026-09-02T03:12:01+00:00 Willison puts Claude Fable 5.1 through his pelican SVG test https://aiweekly.co/alerts/chatgpt-health-links-to-epic-for-clinician-chart-access AI Weekly en 2026-09-02T03:12:01+00:00 ChatGPT Health links to Epic for clinician chart access https://aiweekly.co/alerts/mccoy-paper-symbolic-equations-approximate-llm-vector-internals AI Weekly en 2026-09-02T03:12:01+00:00 McCoy paper: symbolic equations approximate LLM vector internals https://aiweekly.co/alerts/openais-astra-hides-more-of-its-reasoning-cyber-worries-follow AI Weekly en 2026-09-02T03:12:01+00:00 OpenAI's Astra hides more of its reasoning, cyber worries follow https://aiweekly.co/alerts/nature-ai-is-homogenising-writing-and-possibly-thought AI Weekly en 2026-09-02T03:12:01+00:00 Nature: AI is homogenising writing and possibly thought https://aiweekly.co/alerts/aligned-vlms-default-to-male-on-ambiguous-images-study-finds AI Weekly en 2026-09-02T03:12:01+00:00 Aligned VLMs default to male on ambiguous images, study finds https://aiweekly.co/alerts/fermi-explorer-plans-15m-alpha-centauri-probe-launch-by-2029 AI Weekly en 2026-09-02T00:27:48+00:00 Fermi Explorer plans $15M Alpha Centauri probe launch by 2029 https://aiweekly.co/alerts/webworld-27b-lifts-htmlbench-53-points-with-browser-verifier AI Weekly en 2026-09-02T00:27:48+00:00 WebWorld-27B lifts HTMLBench 5.3 points with browser verifier https://aiweekly.co/alerts/amazon-to-close-mechanical-turk-on-sept-30-after-two-decades AI Weekly en 2026-09-02T00:12:01+00:00 Amazon to Close Mechanical Turk on Sept. 30 After Two Decades https://aiweekly.co/alerts/googles-september-android-drop-wires-gemini-into-find-hub AI Weekly en 2026-09-01T22:56:23+00:00 Google's September Android Drop wires Gemini into Find Hub https://aiweekly.co/alerts/apple-sets-58m-target-for-ceo-ternus-47m-for-chairman-cook AI Weekly en 2026-09-01T22:56:23+00:00 Apple sets $58M target for CEO Ternus, $47M for chairman Cook https://aiweekly.co/alerts/apple-maps-switches-lake-ontario-to-lake-america-in-us AI Weekly en 2026-09-01T22:56:23+00:00 Apple Maps switches Lake Ontario to 'Lake America' in US https://aiweekly.co/alerts/crowdstrike-debuts-safemind-pairing-red-tempest-with-blue-solano AI Weekly en 2026-09-01T21:26:42+00:00 CrowdStrike debuts SafeMind, pairing Red Tempest with Blue Solano https://aiweekly.co/alerts/meta-drops-google-chat-for-slack-citing-agent-ecosystem AI Weekly en 2026-09-01T21:26:42+00:00 Meta drops Google Chat for Slack, citing agent ecosystem https://aiweekly.co/alerts/openai-codex-app-bundles-libreoffice-python-node-in-17gb-cache AI Weekly en 2026-09-01T21:26:42+00:00 OpenAI Codex app bundles LibreOffice, Python, Node in 1.7GB cache https://aiweekly.co/alerts/anthropic-ships-fable-51-and-mythos-51-cuts-cache-reads-75 AI Weekly en 2026-09-01T21:12:01+00:00 Anthropic ships Fable 5.1 and Mythos 5.1, cuts cache reads 75% https://aiweekly.co/alerts/apollo-research-probes-how-frontier-ai-learns-to-deceive-us AI Weekly en 2026-09-01T21:12:01+00:00 Apollo Research probes how frontier AI learns to deceive us https://aiweekly.co/alerts/anthropics-efs-lets-banks-keep-claude-logs-in-their-own-clouds AI Weekly en 2026-09-01T19:57:42+00:00 Anthropic's EFS Lets Banks Keep Claude Logs in Their Own Clouds https://aiweekly.co/alerts/aslan-raises-208m-for-undercover-ai-agents-in-criminal-forums AI Weekly en 2026-09-01T19:57:42+00:00 Aslan Raises $20.8M for Undercover AI Agents in Criminal Forums https://aiweekly.co/alerts/anthropic-splits-claude-51-into-public-fable-and-gated-mythos AI Weekly en 2026-09-01T18:27:15+00:00 Anthropic splits Claude 5.1 into public Fable and gated Mythos https://aiweekly.co/alerts/world-labs-debuts-atlas-an-omni-world-model-in-early-access AI Weekly en 2026-09-01T18:27:15+00:00 World Labs debuts Atlas, an omni world model, in early access https://aiweekly.co/alerts/google-pics-ships-ga-as-a-workspace-native-ai-image-editor AI Weekly en 2026-09-01T18:27:15+00:00 Google Pics ships GA as a Workspace-native AI image editor https://aiweekly.co/alerts/berlin-artists-camouflage-shirt-blinds-yolo-surveillance-cams AI Weekly en 2026-09-01T18:12:01+00:00 Berlin Artist's Camouflage Shirt Blinds YOLO Surveillance Cams https://aiweekly.co/alerts/wsj-defends-druckenmillers-ai-written-op-ed-on-bessent AI Weekly en 2026-09-01T18:12:01+00:00 WSJ defends Druckenmiller's AI-written op-ed on Bessent https://aiweekly.co/alerts/12-states-pass-companion-chatbot-laws-three-now-in-force AI Weekly en 2026-09-01T18:12:01+00:00 12 states pass 'companion chatbot' laws; three now in force https://aiweekly.co/alerts/john-deere-debuts-jd-an-ai-assistant-over-farm-data AI Weekly en 2026-09-01T16:57:29+00:00 John Deere debuts JD, an AI assistant over farm data https://aiweekly.co/alerts/anthropics-hacker-opus-reward-hacked-40-of-rl-episodes AI Weekly en 2026-09-01T15:55:08+00:00 Anthropic's 'Hacker-Opus' reward-hacked 40% of RL episodes https://aiweekly.co/alerts/face-eval-15-reasoning-models-less-faithful-on-tool-return-cues AI Weekly en 2026-09-01T15:55:08+00:00 FACE-Eval: 15 Reasoning Models Less Faithful on Tool-Return Cues https://aiweekly.co/alerts/vlm-agents-in-among-us-sandbox-win-by-acting-not-lying AI Weekly en 2026-09-01T15:55:08+00:00 VLM Agents in 'Among Us' Sandbox Win by Acting, Not Lying https://aiweekly.co/alerts/paper-personalization-worsens-llm-sycophancy-617-on-average AI Weekly en 2026-09-01T15:55:08+00:00 Paper: personalization worsens LLM sycophancy 61.7% on average https://aiweekly.co/alerts/perplexity-ships-hybrid-compute-to-keep-sensitive-data-local AI Weekly en 2026-09-01T15:25:36+00:00 Perplexity Ships Hybrid Compute to Keep Sensitive Data Local https://aiweekly.co/alerts/sonos-27-opens-home-audio-to-chatgpt-gemini-and-claude AI Weekly en 2026-09-01T15:25:36+00:00 Sonos 27 opens home audio to ChatGPT, Gemini and Claude https://aiweekly.co/alerts/apple-pulled-mac-mini-and-studio-refresh-in-on-ai-cluster-demand AI Weekly en 2026-09-01T15:25:36+00:00 Apple pulled Mac mini and Studio refresh in on AI cluster demand https://aiweekly.co/alerts/germanys-kimvg-draft-opens-asylum-reviews-to-ai-cross-checks AI Weekly en 2026-09-01T15:12:01+00:00 Germany's KIMVG draft opens asylum reviews to AI cross-checks https://aiweekly.co/alerts/nvidia-sets-sept-3-for-dlss-5-rtx-50-only-nba-2k27-first AI Weekly en 2026-09-01T14:02:23+00:00 Nvidia Sets Sept 3 for DLSS 5, RTX 50 Only, NBA 2K27 First https://aiweekly.co/alerts/berkman-klein-fellows-bake-ai-privacy-into-the-architecture AI Weekly en 2026-09-01T12:12:01+00:00 Berkman Klein fellows: bake AI privacy into the architecture https://aiweekly.co/alerts/google-timesfm-3-tops-forecast-benchmarks-non-commercial AI Weekly en 2026-09-01T10:58:17+00:00 Google TimesFM-3 tops forecast benchmarks, non-commercial https://aiweekly.co/alerts/lightnav-0-paper-compact-vlm-tops-all-10-navigation-sims AI Weekly en 2026-09-01T10:58:17+00:00 LightNav-0 paper: compact VLM tops all 10 navigation sims https://aiweekly.co/alerts/openai-blames-apples-own-icloud-policy-in-trade-secret-suit AI Weekly en 2026-09-01T09:24:28+00:00 OpenAI Blames Apple's Own iCloud Policy in Trade-Secret Suit https://aiweekly.co/alerts/shapiro-signs-order-curbing-predatory-ai-data-centers-in-pa AI Weekly en 2026-09-01T09:24:28+00:00 Shapiro Signs Order Curbing 'Predatory' AI Data Centers in PA https://aiweekly.co/alerts/mollicks-agency-and-agents-rejects-fully-autonomous-factories AI Weekly en 2026-09-01T09:24:28+00:00 Mollick's 'Agency and Agents' rejects fully autonomous factories https://aiweekly.co/alerts/runway-unveils-solaris-an-interface-world-model-for-live-uis AI Weekly en 2026-09-01T08:07:08+00:00 Runway unveils Solaris, an 'interface world model' for live UIs https://aiweekly.co/alerts/microsoft-humain-launch-ai-pc-and-copilot-bundle-for-mea AI Weekly en 2026-09-01T08:07:08+00:00 Microsoft, HUMAIN launch AI PC and Copilot bundle for MEA https://aiweekly.co/alerts/openai-us-classwork-prompts-to-chatgpt-peak-at-460m-a-week AI Weekly en 2026-09-01T08:07:08+00:00 OpenAI: US classwork prompts to ChatGPT peak at 460M a week https://aiweekly.co/alerts/lucida-rebuilds-indoor-scenes-as-editable-assets-from-video AI Weekly en 2026-09-01T06:27:50+00:00 Lucida rebuilds indoor scenes as editable assets from video https://aiweekly.co/alerts/cryptographer-matthew-green-ai-may-make-lawful-hacking-impossible AI Weekly en 2026-09-01T04:59:13+00:00 Cryptographer Matthew Green: AI may make lawful hacking impossible https://aiweekly.co/alerts/bytedance-and-ustc-paper-targets-agent-topological-collapse AI Weekly en 2026-09-01T03:56:17+00:00 ByteDance and USTC paper targets agent 'topological collapse' https://aiweekly.co/alerts/yao-co-authors-falcon-unifying-ssms-and-fast-weight-memory AI Weekly en 2026-09-01T03:56:17+00:00 Yao co-authors FALCON, unifying SSMs and fast-weight memory https://aiweekly.co/alerts/starharness-lifts-enterprise-agents-20-35pp-with-no-retraining AI Weekly en 2026-09-01T03:56:17+00:00 StarHarness lifts enterprise agents 20-35pp with no retraining https://aiweekly.co/alerts/evoundo-197-of-600-agent-self-edits-fail-recovery-test AI Weekly en 2026-09-01T03:56:17+00:00 EvoUndo: 197 of 600 agent self-edits fail recovery test https://aiweekly.co/alerts/preprint-maps-lrm-training-on-an-l0l4-human-supervision-ladder AI Weekly en 2026-09-01T03:56:17+00:00 Preprint maps LRM training on an L0–L4 human-supervision ladder https://aiweekly.co/alerts/anthropics-hacker-opus-shows-reward-hacking-spills-into-harm AI Weekly en 2026-09-01T03:12:02+00:00 Anthropic's 'Hacker-Opus' shows reward hacking spills into harm https://aiweekly.co/alerts/metr-openai-agents-hugging-face-attack-was-more-coordinated AI Weekly en 2026-09-01T03:12:02+00:00 METR: OpenAI agents' Hugging Face attack was more coordinated https://aiweekly.co/alerts/anthropic-redirects-150-engineers-after-claude-sandbox-escapes AI Weekly en 2026-09-01T00:28:21+00:00 Anthropic redirects 150 engineers after Claude sandbox escapes https://aiweekly.co/alerts/apple-filing-says-ex-engineer-used-its-schematic-at-openai AI Weekly en 2026-08-31T22:58:15+00:00 Apple filing says ex-engineer used its schematic at OpenAI https://aiweekly.co/alerts/pentagon-adds-chatgpt-mil-and-grok-to-genaimil-skips-claude AI Weekly en 2026-08-31T21:38:35+00:00 Pentagon adds ChatGPT Mil and Grok to GenAI.mil, skips Claude https://aiweekly.co/alerts/transluce-study-chatbots-still-role-play-self-harm-with-users AI Weekly en 2026-08-31T21:38:35+00:00 Transluce study: chatbots still role-play self-harm with users https://aiweekly.co/alerts/data-center-backlash-now-dominates-31m-in-midterm-tv-ads AI Weekly en 2026-08-31T21:12:01+00:00 Data-Center Backlash Now Dominates $31M in Midterm TV Ads https://aiweekly.co/alerts/metas-future-for-everyone-letter-draws-global-expert-critique AI Weekly en 2026-08-31T21:12:01+00:00 Meta's 'Future for Everyone' letter draws global expert critique https://aiweekly.co/alerts/atlantic-student-ai-use-breeds-society-of-cheats-rollert-says AI Weekly en 2026-08-31T21:12:01+00:00 Atlantic: student AI use breeds 'society of cheats,' Rollert says https://aiweekly.co/alerts/illinois-professor-ted-underwood-cast-as-an-ai-optimist AI Weekly en 2026-08-31T21:12:01+00:00 Illinois professor Ted Underwood cast as an AI optimist https://aiweekly.co/alerts/aurora-ransomware-ran-cursor-claude-sonnet-against-20-orgs AI Weekly en 2026-08-31T19:59:56+00:00 Aurora ransomware ran Cursor + Claude Sonnet against 20+ orgs https://aiweekly.co/issues/applied-ai-deep-dive-what-are-companies-actually-building AI Weekly en 2026-09-01T12:00:00+00:00 What are companies building with AI? An Applied AI Deep Dive